The concrete illustration of the proposal is not provided here.

We have conducted a number of on-line experiments on issues pertaining to:

Bound variable anaphora in
SOV in Japanese
OSV in Japanese
Resumption in OSV in Japanese
Long-distance OSV in Japanese
Local disjointness and OSV in Japanese

Preliminary experiments address:
So- vs. a-NPs as the bindee
Split antecedence in Japanese

and have obtained confirmed predicted schematic asymmetries in line with many of the predicted schematic asymmetries, which will be reported in my forthcoming works (and are reported in works under review).

We have also conducted a number of on-line experiments, whose results clearly dis-confirm *Schema-based predictions we obtain on the basis of language-particular hypotheses regarding so-called local anaphors in Japanese and show many other alleged (and widely accepted) generalizations in Japanese to be invalid. A very small portion of such results are reported in "Hypothesis testing in generative grammar: Evaluation of predicted schematic asymmetries" Journal of Japanese Linguistics 26, a slightly altered version of which is available here. See also the postings under Generalizations [42432] "Alleged generalizations in Japanese" for some more relevant remarks.

I understand that one finds it difficult to appreciate the methodology being proposed and pursued in my current works, thinking that the proposed methodology is too rigorous and too strict in order for one to realistically carry out one's research (and to be able to publish things and get jobs) unless one sees exactly how predicted schematic asymmetries get deduced and confirmed predicted schematic asymmetires have been obtained in actual experiments. I hope it will not be long before works that contain some of the relevant illustration will be published.

Meanwhile, Emi Mukai's "Bound Variable Construal with a 'Discontinuous' Binder" (WAFL 7) contains a small portion of what will be included in her forthcoming USC dissertation, and it gives you an idea about how the methodology for language faculty science advocated in Hoji 2010 (JJL) and elsewhere can be rigorously applied to the study of a certain "phenomena" pertaining to (or including) so-called floating numerals in Japanese. A draft version of the paper is available here.

Hoji 2006 "Assessing Competing Analyses: Two Hypotheses about 'Scrambling' in Japanese," available here also contains some relevant empirical materials although the conceptual and methodological articulation there is much less satisfactory than in what is stated in the postings under Methodology [42404].

The main proposal in Hoji 2009 contains the thesis in (10).

(10)If we want to discover the properties of the Computational System that is hypothesized to be at the center of the language faculty, what I call a confirmed schematic asymmetry should be considered as the minimal unit of 'facts' for such research; see also (58) below.

Much more emphasis is placed in my current work on confirmed predicted schematic asymmetries than on confirmed schematic asymmetries. The former is necessarily deduced from universal and language-particular hypotheses. But attaining the latter is not an easy task, but that would be a minimal requirement for ensuring testability.

In evaluating one's empirical claim, it is useful to distinguish two types of "data," whether is it a confirmed predicted schematic asymmetry or a confirmed predicted schematic asymmetry. One type is what has led one to one's proposal and the other is what is newly predicted by the proposal. The distinction is similar to the distinction made in:

"There is also a more subtle problem.
When you have put a lot of ideas together to make an elaborate
theory, you want to make sure, when explaining what it fits, that
those things it fits are not just the things that gave you the idea
for the theory; but that the finished theory makes something else
come out right, in addition." (From "Cargo Cult Science" by Richard Feynman, found here, for example (as of March 11, 2012)).

Answers we hear to the question about testability, i.e., the question of how one's proposal can be shown to be "wrong," typically do not distinguish the two "types" of data. While we can talk about testability with respect to how valid the accepted empirical generalizations are, it must be recognized that their validity is something basic and what should really be at stake is how the newly made predictions are borne out. If a given proposal is based on alleged generalizations that can easily be, or have already been, shown to be invalid, however, it is not clear what significance can be assigned to the results of an experiment that tests a new prediction that one has made under hypotheses that presuppose the validity of such generalizations; see some remarks under Generalizations [42423].

In order to proceed with our investigation of the language faculty with the general scientific method in (5) [=the hypothetico-deductive method, HH], we must ensure the following, at least to a minimally satisfactory degree.

(6)a.It is possible to compute the consequences of the "guess."
b.It is possible to determine whether or not the consequences of the "guess" agree with the observations and/or the experimental results.

(6a) and (6b) can in turn be ensured only if (7) and (8) hold, respectively.

(7)The "guess" is part of, or is related to, a larger deductive system.

(8)The consequences of the "guess" are related, ultimately, to something 'observable/measurable'.

Furthermore, we should like to ensure (9) as best as we can.

(9)Disagreement between the consequences of the "guess" on the one hand and the observations and/or the experimental results on the other could lead us to learn something about the language faculty.

The proposal put forth in Hoji 2009 is an attempt to ensure these.

My current work share most of the above. Both Hoji 2009 and my current work emphasizes the importance of:

-deducibility
-testability, reproducibility
-maximizing the significance of experimental results

(9) above was addressed in Hoji 2009 under the "Maximizing our chances of learning from errors" heuristics, but that heuristic is understood now as being subsumed under the "Maximize the significance of experimental results" heuristics.

1. Generative grammar as the study of the language faculty^FN2
FN2: What is contrasted here is "generative grammar as the study of language(s)."

What is intended by Chomsky all along, if not practiced quite as such, is that generative grammar is indeed language faculty science, with the understanding that the language faculty manifests itself as I-language, which in turn consists of the Computational System and the mental lexicon.

See Chomsky's (1986) Knowledge of Language (pp. 25-26, for example) for remarks on "E-language" as an "artifact."

In part due to the historical accidents regarding the context in which Chomsky's Syntactic Structures was prepared and published, many practitioners seem to take "E-language" as the object of inquiry for generative grammar. See the first few pages of Chomsky 1975 The Logical Structure of Linguistic Theory, Introduction for the "historical accidents" intended here.

The fist page of chapter 2 of Syntactic Structures states, "The fundamental aim in the linguistic analysis of a language L is to separate the grammatical sequences which are the sentences of L from the ungrammatical sequences which are not sentences of L and to study the structure of the grammatical sequences. The grammar of will thus be a device which generates all the grammatical sequences of L and none of the ungrammatical ones."

Chomsky distinction between I-language and E-language is his attempt to clarify (again) the goal of his research program, in light of what had gone on in the field, in my view.

As I state in Methodology [42415], I believe that, "unless we start accumulating results in language faculty science based on research that rigorously pursues testability, the research program initiated by Chomsky in the mid-1950s will most likely remain to be regarded as a metaphysical speculation, at least by those outside the field" and my current work "is an attempt to articulate how it is possible to pursue language faculty science as an exact science. It provides a conceptual basis for how that is possible in principle and empirical illustration of how that has actually been done."

This is a somewhat non-technical elaboration of the proposal put forth in my book manuscript A Foundation of Generative Grammar as an Empirical Science (henceforth Hoji 2009) and its general significance and implications; it is meant to be an introduction to Hoji 2009.

I am preparing a book that is substantially different from Hoji 2009. The "new book" is called Language Faculty Science: how it becomes an exact science―a proposal and illustration and a draft of its chapter 1 "Chapter 1: Summary and Outline" is provided under Methodology [42404] "The current articulation of the methodology of language faculty science as an exact science." It is not clear what I will do with Hoji 2009, which is a completed book manuscript.

Under this, I will try to post comments on the handout, clarifying some issues in light of my current understanding.

Hoji 2006 is also available here.

One of the main points of Hoji 2006 is that Saito's (2003) use of so as the intended bindee in bound variable anaphora results in the impossibility to attain further testability with regard to local disjointness effects (i.e., the effects of so-called Principle B of the Binding Theory), because so by itself cannot be used as an argument; see Hoji 2006: 4.1.4.

Under this thread, I will post errors found in the JJL paper "Hypothesis testing in generative grammar: Evaluation of predicted schematic asymmetries," available here.

[The font formatting is lost here; and I am not concerned with that.]

Without g(a, b), we do not have two okExamples corresponding to the *Example. Unlike the case with g(a, b)
considered above, we cannot therefore, in principle, have a reasonable ground for excluding the complete unacceptability of a
*Example to "parsing" difficulty or the unnaturalness of the interpretation of the entire a.

excluding
==>
attributing

(I think I was going with "excluding the possibility of xxx" at one point and was not able to completely fix the sentences here...)

[Please ignore font formatting.]

(It is, in principle,
possible to have such a preliminary experiment as long as g(a, b) is based not only on a universal condition but also on a
language-particular condition because an experiment on the language-particular condition can serve as a preliminary
experiment for the one that involves the universal condition as well as the language-particular condition.

1. A parenthesis at the end is missing. (This is in part a reflection of the fact that JJL does not have a copy editor. We will find more errors like this in the future, I am afraid...)

2. an experiment on the language-particular condition can serve as a preliminary
experiment for the one that involves the universal condition as well as the language-particular condition

This is not quite right. What I had in mind was something like the following: Experiments on the a vs. so distinction and those on split antecedence deal with Japanese-particular, and hence language-particular, hypotheses, i.e., lexical hypotheses distinguishing between a-NPs and so-NPs in a crucial way with respect to their (in)ability to be B of BVA(A, B), regardless of the type of BVA (FD-based, co-I-based, or quirky-binding-based) and the lexical hypothesis about the singular-denoting property of so-itu and so-ko while experiments on the (un)availability of FD-based BVA deal with both universal hypotheses and language-particular hypotheses. We may be able to say that experiments on split antecedence are strictly language-particular, but those on a vs. so (that we have been conducting) are based on not only language-particular hypotheses but also universal hypotheses.

In the experiments on a vs. so (that we have been conducting), i.e., EPSA [10], the universal structural condition is satisfied both in Schema A and in Schema B but Schema A does, but Schema B does not, satisfy the lexical condition for BVA. In the experiments on the LF structural condition on BVA, i.e., EPSA [1], on the other hand, the lexical condition is satisfied both in Schema A and Schema B, but the the universal structural condition is satisfied only in Schema A, but not in Schema B.

So, the exposition above in the JJL paper is a bit sloppy.

The handout was slightly revised on 9/9/2009.

The handout was further revised on 12/4/2009.

Handout #1 in my Syntax Seminar this semester, which can be downloaded here, is a substantially revised version of the handouts used for the talks in Kyoto University and the one at Kyushu University in May and for the first talk at Tohoku University and for the talk at Keio University in June. It serves as an introduction to my book under revision A Foundation of Generative Grammar as an Empirical Science.

The handout was slightly revised on 9/9/2009.

The handout for my Keio talk:
June 17 2009
Keio University
"Generative Grammar as an Empirical Science: its goal and how it can be pursued―with some historical reflections" is here: Keio Talk

The "Examples-only" handout, only a very small portion of which was made use of during my presentation at Keio, is here: Keio Examples

The handout used for my Tohokudai talk:
June 12, 2009
Tohoku University
"Toward A Scientific Study of the Language Faculty: a proposal and implications" is here: Tohoku Talk1

I gave two more talks at Tohoku University, one on 6/13 and the other on 6/15. The Keio talk is a revised version of the Tohoku talk given on 6/15 (henceforth, Tohoku Talk3) although the Keio handout contains some materials from the Tohoku talk on 6/13 (henceforth Tohoku Talk1).

Tohoku Talk1, which was based on the talks I gave at Kyoto University and Kyuusyu University in May 2009, was meant to be a presentation of some of the core aspects of the proposal in my book manuscript A Foundation of Generative Grammar as an Empirical Science (under review). Tohoku Talk3 was meant to address more general issues surrounding what is addressed in Talk1 without getting into the details.

During the presentation at Keio, I skipped many of the materials on the handout. Those who are interested in the core proposal covered in Tohoku Talk1 might want to take a look at "Tohoku Talk1" as well; see above.

Under this thread, I will try to post remarks on the Tohoku and Keio talks, including some of the questions and comments during the presentations and what I think went wrong.

　　　Properties of the Computational System can be best detected when we observe 'phenomena' where something has to be the way it is without any conceivable communicative or pragmatic reasons. And that is why researchers have focused on 'phenomena' of various forms of 'agreement' (e.g., he is sick … but not *he are sick) and cases where certain linguistic form must occur in some designated position in relation to other elements in the sentence in which it occurs (e.g., I wonder what John bought but not *I wonder John bought what). The forms prefixed by "*" are judged to be unacceptable by native speakers of English and their judgments are quite robust and uniform despite the fact that nothing seems to be terribly wrong with such forms from a communicative perspective – in fact when a non-native speaker makes such an utterance, the native speaker would most likely understand what is intended.
　　　If a language does not have what is responsible for such 'agreement' or 'obligatory (dis)placement of elements', as I maintain is the case for Japanese, following Fukui 1986, we must look elsewhere in the language to find a reflection of the Computational System, and that is the area that has been called the interface between the language faculty and the faculty of mind that is responsible for interpreting the output of the Computational System. As briefly noted in [29940], the speaker's judgments (i.e., performance) is bound to be affected by factors that are independent of the Computational System, whether it is his/her introspective judgments on the acceptability of a given sentence (under a specified interpretation) or some other manifestation of linguistic performance). By adopting the thesis that what the Computational System fails to give rise to never has the chance to be judged acceptable, can we have a hope of emerging out of the mud of this inherent difficulty with which we are faced. Suppose we proceed to discover, on the basis of the hypothesis testing that has been suggested in Hoji 2006 and is currently further developed and articulated, that some properties of Japanese are indeed due to the Computational System. We will have then learned that such properties are due to the Computational System but not due to what is responsible for 'agreement' or 'obligatory (dis)placement of elements'. It has been a rather common research practice in recent years to assume that all the properties that arguably stem from the Computational System are due to what is responsible for 'agreement' or 'obligatory (dis)placement of elements'. The general research program in which the Kyodai lectures are embedded should therefore play a corrective function to such simple-minded approaches to the properties of the Computational System. Research on a language like English, where we seem to observe two types of reflections of the Computational System, is faced with an inherent difficulty in teasing apart the two. The general significance of research on a language like Japanese thus lies in the prospect of it providing new insight that is difficult to come by from research on a language like English.

What follows is given in "What's New," dated 1/13/2007. I will make some related remarks under this posting.

1/13/2007
The following is posted at a Kyodai website (http://www.hmn.bun.kyoto-u.ac.jp/langlogic/index.html).
For further discussion, please visit the other Discussion boards at this HP. I will try to provide remarks directly related to what is given below in the Further Discussion board.

***
Hoji Kyodai Lectures (2/14-2/16/2006)
Assessing Hypotheses in Generative Grammar

A fundamental working hypothesis in generative grammar is the existence of the language faculty, understood as an algorithm whose input is a set of items taken from the mental Lexicon of the speaker and whose output is a pair of mental representations - one underlying 'meaning' and the other 'sounds'. The main goal of generative grammar can thus be understood as demonstrating the existence of such an algorithm and discovering its properties. Construed in this way, it is not language as an external 'object' but the language faculty that constitutes the object of inquiry. In the terms of the distinction made by Chomsky in the 1960s, generative grammar is concerned with competence rather than with performance.
　　The data in actual research activities in generative grammar, however, is based on acceptability judgments on a given sentence, whether they are introspective judgments by the researchers or their informants or observation of various other types of reactions by 'subjects' in an experimental setting. I.e., what we deal with in an attempt to discover the properties of the speaker/hearer's competence (the Computational System) is his/her performance (i.e., language use, in a broad sense). This makes it crucial, in the context of generative grammatical inquiry as construed above, to articulate how we can extract from performance data evidence for a hypothesis about the properties of the Computational System. The absence of a minimally satisfactory articulation of how to do this is likely to lead to a situation where different (groups of) researchers base their proposals about the Computational System on different sets of speaker judgments, collected in a variety of ways, that are not necessarily uniform or robust, being subject to a great deal of fluctuation and variation not only among speakers but also within a single speaker. This makes it difficult to evaluate competing proposals in a reliable and objective manner. Recent debates in leading journals (e.g., Language, Lingua and Natural Language and Linguistic Theory) about what should qualify as data, about the use of introspective judgments as crucial evidence for or against a theory, etc. seem to suggest that we have not yet developed a means to evaluate the empirical bases of hypotheses in generative grammar that is compelling enough to the majority of the practitioners. An evaluation of a given hypothesis thus tends to have an arbitrary aspect to it, influenced by such factors as whether or not the terms and concepts utilized are of a theory currently in fashion and whether or not it endorses the standard view concerning the validity of alleged empirical generalizations, regardless of how much 'repeatability' obtains in regard to the predicted speaker judgments on the crucial sentences.
　　In my lectures, I will try to review some of the efforts over the past decade to overcome the problem just noted, including some articulation of a concrete means to evaluate (not to arrive at) hypotheses in generative grammar, making specific reference to Japanese as its empirical basis for an illustration of its methodological point. The methodological aspects of its results should be applicable to research on other languages, as long as it deals with interpretations that are claimed to be based crucially on properties of the Computational System.
　　When one aims at discovering the properties of the language faculty as construed above, one must recognize the following: the Computational System's yielding something as its output does not guarantee that the speaker finds it (more precisely, finds its surface manifestation) to be acceptable; after all, non-grammatical as well as grammatical factors must contribute to the ultimate acceptability judgment by the speaker on a given sentence form under a specified interpretation (e.g., one's knowledge about the world, one's belief system, and the like). The Computational System's failure to yield something as its output, on the other hand, should necessarily mean that the 'sentence form' corresponding to such a 'failed representation' should be judged unacceptable under the specified interpretation. If something is predicted to be impossible due to the hypothesized formal properties of the Computational System under discussion, how can some pragmatic adjustment save it? Only by taking this point to heart and by putting forth a hypothesis so as to yield a negative prediction (the prediction that something is impossible (under a specified interpretation)), do we have a hope of making generative grammar an empirical science or of making it a progressive research program in the sense of Lakatos 1970.
　　The major concerns underlying the research reported here are:

(i)　a. 　How can we try to ensure and measure progress in what we do in generative grammar?
　　b. 　How can we tell whether or not given intuitions of ours are likely to be a reflection of the Computational System?

　　Not every observation qualifies as something that must be accounted for by a theory about the Computational System; it must first be demonstrated that it is most likely a reflection of the Computational System. And, for the reasons briefly noted above and further elaborated elsewhere (e.g., Hoji 2006), to do so would require the recognition of the significance of negative predictions insofar as the research in question is aimed at demonstrating the existence of and discovering properties of the Computational System, which is hypothesized to be at the core of the language faculty. This is the central methodological claim in the research reported here and the lectures are meant to illustrate its content and how it relates to actual empirical materials.

Day 1 (2/14) (Bungakubu Higashikan Lecture Room 4)
Session 1
How (I think) we should proceed if the aim of our research is to demonstrate the existence of and discover the properties of the Computational System that is hypothesized at the core of the language faculty. I will try to provide a brief overview of how we have proceeded to isolate certain linguistic intuitions that are likely based on the Computational System. Coreference to BVA (bound variable anaphora) and any BVA to FD-based BVA (i.e., BVA that is based on a c-command relation at LF), and how we have come to identify FD-based BVA, what problems we have faced and are still facing in doing so, how we have tested our hypotheses, and how the FD-based-BVA-related discussion has been related to other phenomena (such as FD-sloppy readings and local disjointness effects).

Session 2
How one could fail to do what should be done, according to the discussion in Session 1. I will also address general issues of the failure to pay (serious) attention to negative predictions, falsifiability, repeatability, etc., and its consequences (as observed in the field). I will try to go over some correlations among different types of research practice, which seem to me to stem from different types of research orientations and goals.

Day 2 (2/15) (Bungakubu Shinkan Lecture Room 1)
Session 3
Concrete illustration of the methodological points made in Sessions 1 and 2, including how to assess hypotheses in generative grammar -- given that the research goal is as described in Sessions 1 and 2 -- by making reference to two competing analyses of the so-called scrambling construction in Japanese (Ueyama 1998 and Saito 2003). Attempts will also be made to illustrate how one might proceed when one faces apparent disconfirmation of one's negative prediction. The audience will be asked, but not forced, to participate in a series of experiments in class and also on-line afterwards.

Session 4
Questions and comments from the audience, their articulation and responses.
Potential topics to cover, if the time remains, include: (i) attempts to 'apply' the binding theory to Japanese, (ii) quantifier scope, (iii) negation, and (iv) ellipsis, and (v) the Subjacency effects and anaphoric dependency. Depending upon how the discussion goes, I may go over some of these topics in relation to the methodological and empirical points made in Sessions 1-3.

Day 3 (2/16) (Bungakubu Shinkan Lecture Room 1)
Session 5
The topics to be chosen from what is listed above for Session 4.

Session 6
Concluding remarks. Reiterating the main concerns and the proposals.

Readings:
General: Hoji 2003.
Day One: The handouts by Ueyama and Hoji for the 2006 Hokudai Kagaku tetsugakkai workshop, Hoji 1997/2006, and Hoji 2006.
Day Two: Ueyama 2002, Saito 2003, Hoji et al. 1999.
Day Three: (It is not clear what will actually be covered in the lectures.) Hayashishita 2004: Chap. 2, 2.2.1, Kataoka 2007 (to appear in Gengokenkyuu).

Hoji 1997/2006, 2003, 2006, Hoji et al 1999, Ueyama 2002, and Hayashishita 2004 are available on-line. Please visit http://www.gges.xyz/hoji/research/hp-papers.cgi for the first four, and for Ueyama 2002 and Hayashishita 2004, please go to http://www.gges.xyz/hoji/research/hp-Ayumi.cgi and http://enteroflora.com/linguistics/dissertation.html, respectively. For Kataoka 2007, please email me at hoji@usc.edu.

Hayashishita, J.-R. 2004. Syntactic and Non-Syntactic Scope, Doctoral dissertation. Los Angeles: University of Southern California.
Hoji, H. 1997/2006 "Otagai," Ms. Los Angeles: University of Southern California. [Presented at WCCFL 16, University of Washington, March 1997.] Published in A. Ueyama (ed.), 2006. Theoretical and empirical studies of reference and anaphora-Toward the establishment of generative grammar as an empirical science, Kyushu University, pp. 126-138.
Hoji, H. 2003. "Falsifiability and Repeatability in Generative Grammar: A Case Study of Anaphora and Scope Dependency in Japanese." Lingua 113/4-6, pp. 377-446.
Hoji, H. 2006. "Two Hypotheses about Scrambling in Japanese," in A. Ueyama (ed)., Theoretical and empirical studies of reference and anaphora-Toward the establishment of generative grammar as an empirical science, Kyushu University, pp. 139-185.
Hoji, H., S. Kinsui, Y. Takubo, and A. Ueyama 1999. "Demonstratives, Bound Variables, and Reconstruction Effects." in Proceedings of the Nanzan GLOW, The Second GLOW Meeting in Asia, September 19-22, 1999. pp. 141-158.
Kataoka, K. 2007. "Negをc-統御する不定語+モ," 『言語研究』131 号 (2007年 3 月刊行予定).
Lakatos, I. 1970. "Falsification and Methodology of Scientific Research Programmes," in I. Lakatos and A. Musgrave (eds.), Criticism and the Growth of Knowledge. Cambridge University Press. pp. 91-195.
Saito, M. 2003. "A Derivational Approach to the Interpretation of Scrambling Chains," Lingua 113/4-6, pp.481-518.
Ueyama, A. 1998. Two Types of Dependency. Doctoral dissertation. Los Angeles: University of Southern California. (Distributed by GSIL publications. Los Angeles: University of Southern California.)
Ueyama, A. 2002. "Two Types of Scrambling Constructions," in A. Barss (ed.), Anaphora: A Reference Guide, Blackwell.

Given below is the table of contents of Theories of Anaphora and Aspects of Japanese Syntax (unpublished book manuscript, 1990), which I never got around to revising for publication from the MIT Press, despite the encouragement by the editor (J. Keyser) to revise it as I would find appropriate. At the time, I felt that there was just so much to do to make it satisfactory -- and that was certainly true --, I never got around to forcing myself to face the task. But what has transpired in the field since 1990 now makes me think that it might not have been a bad idea to make its content available to the public. (Some of my colleagues in fact have repeatedly told me that I should have published it, and its publication could have avoided part of the confusion so pervasive in the field at the moment.) Right now, only chapter 5 has been reformatted into a PDF file; but I intend to work on the reformatting of the other chapters. If you are interested in obtaining any of the following chapters, please let me know by email.
***

Chapter 1: Linguistic Theory and the Grammar of Japanese
1.1.　Introduction　1
1.2.　An Introduction to Syntactic Theory　2
1.3.　An Introduction to Issues in Japanese Syntax　8
1.4.　Outline of the Book　27

Chapter 2: Definite NP Anaphora and Japanese Phrase Structure
2.1.　Introduction: Binding Condition C and the VP Node in Japanese　41
2.2.　A Brief History of Binding Condition C　42
2.3.　Condition C in Japanese　47
2.4.　Condition D　52
2.4.1.　Further Evidence for Condition D　56
2.5.　Condition D and C-Command Domains　64
2.6　More Hierarchies　78
2.6.1　Social Titles and Kare　79
2.6.2　Social Titles and Epithets　81
2.7　Inside the NP　82
2.8　The Referential Hierarchy and Binding Theoretic Features　85
2.9　Binding Condition B　92
2.9.1.　Pronouns　92
2.9.2.　Names　102
2.9.3.　Epithets　106
2.9.4.　Social Titles　111
2.9.5.　NP-Internal Structure　117
2.9.6.　A Proposal on Condition B　120
2.10.　Overlapping Coreference　141
2.11.　Social Titles as Descriptions　150
2.12.　Summary　155

Chapter 3: On the Nature of Condition D
3.1.　INTRODUCTION　218
3.2.　A CRUCIAL DIFFERENCE BETWEEN B AND D　220
3.3.　ON THE NATURE OF CONDITION D　234
3.3.1.　A Proposal on Linking　234
3.3.2.　On the Suspension of Condition D　248
3.4.　THE SUSPENSION OF CONDITION D IN ENGLISH　262
3.5.　SOME RELATED ISSUES　271
3.5.1.　Kare v.s. Zibun　271
3.5.2.　Landing sites of Scrambling and NP Movement　299
3.5.2.1.　Levels of Representations　304
3.5.2.1.　"Reconstruction" Effects and Condition D　322
3.6.　ON THE PRAGMATIC LICENSING OF THE SUSPENSION OF D　330
3.7.　SUMMARY　333

Chapter 4: Bound Variable Anaphora in Japanese
4.1.　INTRODUCTION　382
4.2.　BOUND VARIABLE ANAPHORA　386
4.3.　BOUND VARIABLE ANAPHORA IN JAPANESE　388
4.4.　THE SO SYSTEM　392
4.4.1.　D-Linking and Bound Variable Anaphora　392
4.4.2.　No Student　400
4.4.3.　Split Antecedence and the Plurality of Soko/Sore　409
4.4.4.　Soko as a Bound Variable　415
4.5.　"RECONSTRUCTION" AND "PARASITIC GAPS"　422
4.6.　KARE, SORE AND THE JAPANESE DEMONSTRATIVE PARADIGMS　434
4.6.1.　The So-called Overt Pronouns and Sono hito 'that person'　434
4.6.2.　The Japanese Demonstrative System　436
4.6.3.　The A and So Demonstrative Paradigms　438
4.6.4.　Kare and the A system　442
4.7.　THE SO-CALLED OVERT PRONOUNS IN JAPANESE　446
4.8.　DEMONSTRATIVITY AND BOUND VARIABLE CONSTRUAL　449
4.9.　CONDITION B EFFECTS AND BOUND VARIABLE ANAPHORA　455
4.10.　SUMMARY　460

Appendix to Chapter 4

Chapter Five: Sloppy Identity in Japanese
5.1.　INTRODUCTION　509
5.2.　A STANDARD ANALYSIS OF THE SLOPPY/STRICT READINGS　512
5.3.　SOO SU (DO SO)　513
5.3.1.　Kare　514
5.3.2.　Zibun　516
5.3.3.　On the Nature of Soo Su　518
5.4.　STRIPPING　522
5.4.1.　The Sloppy Reading in Stripping　522
5.4.3.　The Stripping in Japanese　526
5.4.3.1.　The Predictions Fail　527
5.4.4.　The Subjacency　530
5.4.4.1.　The Subjacency Violation　538
5.4.5.　The Subjacency in Japanese　540
5.4.5.1.　The Topic Constructions　543
5.4.5.2.　The Cleft Construction　551
5.4.5.3.　The Japanese Stripping Revisited　555
5.4.6.　The Subjacency in the Japanese Stripping　556
5.4.7.　The Sloppy Reading in the Japanese Stripping　559
5.4.7.1.　Sloppy Reading without Bound Variable Construal　570
5.4.7.2.　Summary　581
5.5.　"DEEP AND SURFACE ANAPHORA"　583
5.6.　THE STRUCTURE OF THE STRIPPING CONSTRUCTION　592
5.6.1.　Two Types of Stripping　592
5.6.2.　A Proposal　593
5.6.　THE STRUCTURE OF THE STRIPPING CONSTRUCTION　606
5.6.1.　Two Types of Stripping　606
5.6.2.　A Proposal　607
5.7.　THE SLOPPY READING IN THE JAPANESE COMPARATIVE　620
5.8.　CONDITION B EFFECTS AND SLOPPY READING　624
5.9.　SUMMARY　634

Chapter 6: Coreference, Bound Variable Anaphora and Language Acquisition
6.1.　INTRODUCTION　673
6.2.1.　Coreference　675
6.2.1.1.　Condition B　675
6.2.1.2.　Condition C　680
6.2.1.3.　Condition D　683
6.2.2.　Bound Variable Anaphora　684
6.2.2.1.　Condition B　684
6.2.2.2.　Conditions C and D　686
6.2.3.　A Summary　689
6.3.　REINHART'S "PRAGMATIC ACCOUNT" OF DISJOINT REFERENCE　690
6.4.　AN ACQUISITION PUZZLE　699
6.4.1.　The Acquisition of Condition B　699
6.4.1.1.　Condition B for Coreference and Bound Variable Anaphora　700
6.4.1.2.　Against Reinhart's "Pragmatic Strategies"　702
6.4.1.3.　An Alternative Account: Children's him as deictic　702
6.4.1.4.　Problems　709
6.4.1.5.　Children's Him and the Japanese So　711
6.4.2.　Condition B for Non-Pronouns　720
6.5.　ACCOUNTING FOR THE EFFECTS OF BINDING CONDITIONS FOR COREFERENCE　724
6.5.1.　Condition B　725
6.5.2.　Condition C　733
6.5.3.　Condition D　736
6.6.　Remaining Issues　740
6.7.　Concluding Remarks　741

One might actually argue that the confusing exposition and the poor quality of the research reported in Hoji 1991 is responsible for Elbourne's (2002) 'misrepresentation' of Hoji 1991. I will try to elaborate on this in a posting under this.

Elbourne 2002: chapter 5, p. 179 states:

" ..., then kare and kanozyo should have E-type uses. I do not know of anywhere in the previous literature on these words where this prediction has been tested."

One of the claims of Hoji et al. 2000 is precisely that kare can have an E-type use.

Elbourne, Paul. 2002 Situations and Individuals, Doctoral dissertation, MIT.

The way Elbourne 2002: chapter 5, p. 175 cites (or I should perhaps say, misrepresents) Hoji 1991 is quite remarkable. It in effects states that Hoji (1991) was unaware of the fact that demonstratives (e.g., that logician in English or so-ko in Japanese) can be construed as a bound variable, attributing Hoji 1991 "the puzzling proposal that kare cannot be bound because it is a demonstrative," and continues "He does not take account of the many examples which show demonstratives can in fact be bound."

One of the main empirical points in Hoji 1991 is precisely that so-demonstratives (as opposed to a-demonstratives) can be construed as a bound variable. The puzzling proposal in Hoji 1991 actually has to do with its introduction of the 'degrees of demonstrativity' (as a desperate attempt to account for the difference between so demonstratives and a demonstratives -- and it tries to relate kare to a demonstratives). This is certainly a puzzling and obscure notion, to say the least, and it has (quickly) been abandoned in subsequent works of mine.

I have recently noted the following in "What's new" (9/11/2006).

I have just uploaded Hoji 1990 and Hoji 1991.

Hoji (1990) "On the so-called Overt Pronouns in Japanese and Korean," in E.-J. Baek, ed., Papers from the Seventh International Conference on Korean Linguistics, pp. 61-78.

Hoji (1991) "KARE" in Carol Georgopoulos and Roberta Ishihara, eds., Interdisciplinary Approaches to Language: Essays in Honor of Prof. S.-Y. Kuroda, Reidel, Dordrecht, pp. 287-304.

Hoji 1990 was written after Hoji 1991 "Kare," the final version of which had been written in 1989 but did not appear until 1991.

What has compelled me to upload these old papers is what seems to me to be a persistent refusal in the field to recognize that Japanese does not have anything like a 'pronominal system' that deserves to be so called, in light of the formal properties of the 'potential candidates for 'overt pronouns' in the language. I will try to make postings in the Further Discussion board on this topic.

See also Hoji 1995 (NELS), Hoji et al. 2000 (GLOW), and the references cited in the latter.

Among what I had in mind is a view that can be seen in remarks such as (1).

(1) a. (Elbourne 2002: p. 9)
[Considering "referential and bound uses of pronouns" as "a kind of definite articles"] thus provides a unified semantics for the E-type, bound and referential uses of pronouns, which is surely desirable, since no language makes any lexical or morphological distinction between pronouns used with these allegedly different meanings. (The emphasis by HH.)
b. (Elbourne 2002: p. 18)
As already mentioned, no language shows any lexical or morphological difference between pronouns used as individual variables and pronouns used as definite descriptions.

Elbourne, Paul. 2002 Situations and Individuals, Doctoral dissertation, MIT.

As discussed in the works cited above and in the references therein, there is a clear difference between the so demonstrative and the a demonstrative; it is not impossible for so-NPs to function as a bound variable while it is impossible for a-NPs to do so. If expressions like so-ko, so-re, a-soko, and a-re, all of which can translate English it, are considered as "pronouns," the alleged claim made in (1) is clearly invalid.

Note: Hoji et al. 2000: note 3 states, "-So- in a-soko comes from si in asiko that appeared circa between 800 and 1200, and is unrelated to so in so-ko, although the etymology of this si is not entirely clear." See Hoji 2003 (Lingua): note 10 for more on the basic demonstrative paradigms in Japanese.

Here is Assessing Competing Analyses: Two Hypotheses about 'Scrambling' in Japanese (2006) in Ayumi Ueyama, ed., Theoretical and Empirical Studies of Reference and Anaphora?Toward the establishment of generative grammar as an empirical science, pp. 139-185.

The paper can be considered as a(n) (almost direct) continuation of Hoji 2003 in Lingua and it contains some of the materials discussed in the papers and handouts placed in this board. In particular, some/much of the Mayfest2005-related documents, both empirical and methodological, has/have been incorporated in the paper. The methodological issues addressed in the Major-Object-related documents placed in this board are addressed in this paper, hopefully in a way that is clearer than before.

A slightly revised handout for the Mayfest talk, prepared for my syntax seminar at USC, spring of 2006 has been uploaded.

Here is the syllabus for the course.

Here is the handout used for my presentation at the LSA Workshop on Control and Raising.

A Major Object Analysis of the So-called Raising-to-Object Construction in Japanese

The supplementary handout I used during my presentation is here.

A slightly revised handout for the Mayfest talk, prepared for my syntax seminar at USC, spring of 2006 has been uploaded.

Two Hypotheses about Scrambling in Japanese: Ueyama 2003 vs. Saito 2003

The examples that are made reference to in that handout are in MayFest Examples Handout.

It seems that the bulk of my presentation will be concerned with justifying my position declared in the second paragraph of the abstract, given in [21482], repeated here.

We adopt the following thesis, put forth in Ueyama 1998, according to which the movement operations involved in the OS construction (i.e., sentences of the 'object subject order') in Japanese are not uniform, contrary to the thesis that seems to have been pursued for the past two decades by M. Saito, and a given example of the OS construction in Japanese can correspond to two distinct numerations, and hence two distinct derivations and representations. The OS order can come about as the result of the PF movement of a non-subject NP over the subject, but it can also come about with the 'base-generation' of the non-subject (such as the object) at the sentence-initial position, being related to 'its theta position' through a Predication relation with a lambda predicate containing 'the theta position' as the open position. One might suggest that the former is akin to Heavy NP shift in English, to the extent that it is analyzed as an instance of PF movement, and that the latter is not unlike what happens in the tough sentences in English.

The crucial difference between Ueyama 1998 and Saito 2003 is (i). What is meant by DL in (i) (adopting the term in Ueyama) is 'dislocated phrase', and it can be understood as corresponding to the so-called scrambled phrase.

(i) Ueyama 1998 deduces the impossibility of one DL simultaneously functioning 'as being in an A-position' and 'as being in an A'-position', while Saito 2003 does not. In fact, Saito's (2003) analysis allows of the possibility that a given DL exhibits both A-properties and A'-properties since the formal relation that underlie A-properties and the one that underlie A'-properties can be established/licensed derivationally
Ueyama 1998 thus makes a particular negative prediction while Saito 2003 does not. The bulk of my MayFest presentation will likely be concerned with the demonstration that Ueyama's hypothesis at least is not falsified, if not corroborated, and to the extent such is indeed the case, that would provide (strong) support for Ueyama's theory of the OS construction (i.e., so-called scrambling construction), as opposed to a theory such as Saito's (2003).

I think I will also be able to go over the point alluded to in the last paragraph of the abstract, also repeated here.

One might point to various arguments in the literature for the existence of formal features (strong features, the EPP feature, or the like) in Japanese. However, once we put such proposals to a minimally rigorous empirical test, checking its negative predictions, it is not clear if any such argument/hypothesis/claim remains not to be refuted/falsified/invalidated. We will provide illustration of some concrete examples. One might also wonder how valid the empirical bases are for the claims defended in this presentation, pointing to judgments reported in the literature in support of the hypotheses/claims contrary to the thesis pursued here. Among the key notions in assessing a given hypothesis and in dealing with judgmental fluctuation are falsifiability, corroboration (not in the sense of Popper), and negative predictions, and the recognition that a contrast detected in a minimal pair should be counted as significant only if the negative prediction made by the hypothesis in question is not disconfirmed. I will try to go over what is meant by this and also try to make some remarks on what role can be profitably served by research on Japanese (and other languages that share with Japanese the crucial properties noted above), given the basic correctness of the general thesis pursued here.

But I am afraid that I will not be able to address the issues of clause-boundedness and the Subjacency effects, in any depth. Emi Mukai's recent paper directly addresses those issues (and 'resumption'), and I would most likely have to refer the interested parties to her work, which I plan to upload at my HP, when it has been revised for public circulation.

The main reason for this 'change of plan' is that it seems necessary to illustrate the format of the experiment introduced in the handout posted in [21051] as well as in my abstract for the LSA 2005 Raising and Control Workshop A Major Object Analysis of the So-called Raising-to-Object Construction in Japanese . (A slightly longer version of the abstract is here.) I plan to do so, by comparing Ueyama 1998, 2002 and Saito 2003, and if I do that, I doubt that I will have enough time to discuss the concrete issues regarding clause-boundedness and the Subjacency effects.ted here.

We adopt the following thesis, put forth in Ueyama 1998, according to which the movement operations involved in the OS construction (i.e., sentences of the 'object subject order') in Japanese are not uniform, contrary to the thesis that seems to have been pursued for the past two decades by M. Saito, and a given example of the OS construction in Japanese can correspond to two distinct numerations, and hence two distinct derivations and representations. The OS order can come about as the result of the PF movement of a non-subject NP over the subject, but it can also come about with the 'base-generation' of the non-subject (such as the object) at the sentence-initial position, being related to 'its theta position' through a Predication relation with a lambda predicate containing 'the theta position' as the open position. One might suggest that the former is akin to Heavy NP shift in English, to the extent that it is analyzed as an instance of PF movement, and that the latter is not unlike what happens in the tough sentences in English.

The crucial difference between Ueyama 1998 and Saito 2003 is (i). What is meant by DL in (i) (adopting the term in Ueyama) is 'dislocated phrase', and it can be understood as corresponding to the so-called scrambled phrase.

(i) Ueyama 1998 deduces the impossibility of one DL simultaneously functioning 'as being in an A-position' and 'as being in an A'-position', while Saito 2003 does not. In fact, Saito's (2003) analysis allows of the possibility that a given DL exhibits both A-properties and A'-properties since the formal relation that underlie A-properties and the one that underlie A'-properties can be established/licensed derivationally.

Ueyama 1998 thus makes a particular negative prediction while Saito 2003 does not. The bulk of my MayFest presentation will likely be concerned with the demonstration that Ueyama's hypothesis at least is not falsified, if not corroborated, and to the extent such is indeed the case, that would provide (strong) support for Ueyama's theory of the OS construction (i.e., so-called scrambling construction), as opposed to a theory such as Saito's (2003).

I think I will also be able to go over the point alluded to in the last paragraph of the abstract, also repeated here.

One might point to various arguments in the literature for the existence of formal features (strong features, the EPP feature, or the like) in Japanese. However, once we put such proposals to a minimally rigorous empirical test, checking its negative predictions, it is not clear if any such argument/hypothesis/claim remains not to be refuted/falsified/invalidated. We will provide illustration of some concrete examples. One might also wonder how valid the empirical bases are for the claims defended in this presentation, pointing to judgments reported in the literature in support of the hypotheses/claims contrary to the thesis pursued here. Among the key notions in assessing a given hypothesis and in dealing with judgmental fluctuation are falsifiability, corroboration (not in the sense of Popper), and negative predictions, and the recognition that a contrast detected in a minimal pair should be counted as significant only if the negative prediction made by the hypothesis in question is not disconfirmed. I will try to go over what is meant by this and also try to make some remarks on what role can be profitably served by research on Japanese (and other languages that share with Japanese the crucial properties noted above), given the basic correctness of the general thesis pursued here.

But I am afraid that I will not be able to address the issues of clause-boundedness and the Subjacency effects, in any depth. Emi Mukai's recent paper directly addresses those issues, and I would most likely have to refer the interested parties to her work, which I plan to upload at my HP, when it has been revised for public circulation.

If you can read Japanese, you might want to take a look at a recent paper by Ayumi Ueyama -- 上山あゆみ (2006) 「経験科学としての生成文法―文法性と容認可能性―」, 九州大学言語学論集 25/26 (Ueyama, Ayumi (2006) "Generative Grammar as an Empirical Science---Grammaticality and Acceptability---", Kyushu University Papers in Linguistics, vol.25/26.)-- which directly addresses the methodological point alluded to in my Mayfest talk and elsewhere.

If you find [18314] unclear and if you would like to know what was intended there, please take a look at the Kyodai handout (under [20194]) and the Mayfest abstract (under [21482]), and the other-related postings, all of which were made rather recently.

As of April, 2005, I am using again corroboration, but with the qualification not in Popper's sense, thinking that it is perhaps necessary for us to establish this use of corroboration in generative grammar as an empirical science with progress in mind.

So, it is crucial that we use a QP for which BVA(QP, B) is possible only on the basis of LF c-command if we are conducting tests that make crucial reference to (LF) c-command. [And also the term for B, HH. 4/2/2005]. The failure to do so has resulted in a great deal of judgmental instability, to put it somewhat mildly, in regard to the alleged generalizations discussed in the literature for the past two decades, not only in the area of BVA but also in the area of quantifier scope. The same holds true also in the case of the generalization illustrated by (16) in the WECOL handout and the various paradigms of resumption examples discussed there; and in this sense, the paradigms in the WECOL handout can be understood as constituting strong confirming evidence for the theory of anaphoric relations proposed in Ueyama 1998, in which different types of BVA are teased apart carefully.

Suppose we conduct experiments on the (un)availability of BVA, on the basis of (i) the theory of anaphoric relations in Ueyama 1998, in which such distinctions are made among different sources of BVA, and (ii) what seems to be assumed in Saito 2003, for example, in which such distinctions are not made.

Note that Ueyama (1998) argues that we must be careful in choosing both the 'binder' and the 'bindee' if we are to consider the LF-command-based BVA. A typical 'binder'-'bindee' pair used in Saito 2003 is (i).

(i) dono N-mo ... sono-N
E.g., Dono hon-mo ... sono tyosya-CM ...
(Lit) which book-also ... the author-CM (CM being a case marker)

What is intended in Saito 2003 is to take sono as 'its' with 'it' being bound by the quantifier, and the rough translation for the second line in (i) is (intended to be) "every book ... its author."

But 'binder'-'bindee' pairs like (i) are precisely among those Ueyama (1998) states should not be used in our experiments intended to probe into LF-command-based intuitions -- well, the point re. sono tyosya 'the/its author' might not be made directly in the thesis but it is not difficult to extrapolate it from the discussion there. I am not going to get into the reasons here; but you can check the discussion in Ueyama 1998 on 'co-I-indexation' and also Appendix D. Ueyama 1998 provides Lists of Abbreviations and Terms, which functions a bit like Index; so you can get to the relevant parts of the thesis fairly easily (well, much more easily than otherwise).

So, as expected, the results of the two experiments noted above have yielded substantially different results (clear falsification in one case and corroboration in the other, by the 'criteria' noted in [21482]). I will try to go over that at my Mayfest presentation if the materials fit in the allotted time.

I should also note that only two speakers among the 23 speakers judged all of the Japanese analogues of (1b), (2b), and (3b) as "-2," in sharp contrast with the result that six or more speakers (among the nine) judged all of English (1b), (2b), and (3b) as "-2."

I asked some of the informants to judge the sentences again, without looking at their judgments that they had previously reported, including those two speakers alluded to above. Some gave slightly lower scores than before and others higher scores, but not in any way that can be understood as being systematic. Of the two speakers referred to above, one gave 0, +1, and +1 to the three examples in question, instead of -2, -2, -2, while the other informant gave -2, -2 and -2, again.

If one insisted on being interested in the 'grammar' of a speaker who takes zibunzisin as a local anaphor, one would presumably be examining only one out of the 23 speakers... Even that one speaker, upon being provided with some pragmatic context, I suspect, will start accepting the relevant examples, and that, I would also suspect, would contrast sharply with what would happen with English sentences with reflexives.

What need to be borne in mind is, again, that if something is predicted to be impossible for a grammatical reason, no pragmatic adjustment should save it. And that is how we can hope to attain falsifiability in what we do; see my Mayfest abstract for brief remarks on the additional 'requirement' (for attaining corroboration).

(i) the locality of so-called numeral floating quantifiers

The unexpected acceptability (for many speakers) of what is claimed to be unacceptable is typically attributed to some extra factor, such as focus. To the extent that the notion 'focus' can be made explicit and the relevant claim is made falsifiable, such an attempt to save the alleged generalization turns out to be of not much help. We checked this while preparing the WCCFL paper (UCDavis) although we did not mention it in the paper since we did not find it necessary to do so; the paper shows that even if the focus-based attempt to save the generalization were successful, a particular implementation of the idea such as the one found in Miyagawa and Arikawa (2003, the UCLA 12/2004 handout) -- see the WCCFL paper by Hoji&Ishii -- makes a wrong negative prediction and hence their hypothesis is falsified. If one wished to propose a different implementation, one should try that. Once that has been done, its empirical consequences can be tested in regard to the negative predictions it makes.